RESEARCH ARTICLE Open Access

Multilevel analysis of factors associatedwith unmet need for family planningamong Malawian womenOwen Nkoka1,2*, Watanja M. Mphande3, Peter A. M. Ntenda1,4, Edith B. Milanzi1, Victor Kanje1 and Shiaau J. G. Guo2

Abstract

Background: Malawi has a high fertility rate which is also characterized by a relatively high prevalence of unmetneed for contraception. However, little is known about the influence of individual- and community- level characteristicson unmet need in Malawi. This study examined the individual- and community- level factors associated with unmetneed for family planning (FP) among Malawian women.

Methods: Data from the 2015–16 Malawi demographic and health survey were used to analyze 15, 931 women. Theassociation between individual- and community- level factors and unmet need was assessed using multilevel binarylogistic regression models.

Results: The prevalence of total unmet need was 21.0%. Women aged ≥35 years were more likely to have total unmetneed [adjusted odds ratio (aOR) = 1.19, 95% confidence interval (CI) = 1.04–1.35] compared with those aged 15–24years. Women who were married [aOR = 0.41, 95% CI = 0.35–0.48], and those employed [aOR = 0.78, 95% CI = 0.71–0.85] were associated with less likelihood of having total unmet need compared with unmarried, and unemployedwomen, respectively. At community-level, women from communities with a high percentage of women from richhouseholds [aOR = 0.81, 95% CI = 0.67–0.96], and those from communities with a middle and high percentage ofeducated women [aOR = 0.86, 95% CI = 0.76–0.96 and aOR = 0.81, 95% CI = 0.70–0.93, respectively] were less likely tohave total unmet need for FP compared with those from communities with low percentages of rich and educatedwomen, respectively. The proportional change in variance showed that about 36.0% of total variations in the odds ofunmet need across the communities were explained by both individual- and community-level factors. Moreover, theintraclass correlation showed that about 3.0% of the total variation remained unexplained even after controlling forboth individual- and community-level factors.

Conclusion: Both individual- and community- level factors influenced unmet need for FP in Malawi. Public healthpractitioners should conduct community profiling and consider individual and community factors when designing FPprograms.

Keywords: Unmet need, Family planning, Multilevel, Malawi

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to thedata made available in this article, unless otherwise stated in a credit line to the data.

* Correspondence: onkoka@gmail.com1Institute for Health Research and Communication (IHRC), P.O Box 1958,Lilongwe, Malawi2School of Public Health, College of Public Health, Taipei Medical University,250 Wuxing Street, Xinyi Taipei, Taiwan 110Full list of author information is available at the end of the article

Nkoka et al. BMC Public Health (2020) 20:705 https://doi.org/10.1186/s12889-020-08885-1

BackgroundThe use of contraceptives to regulate fertility either forchild spacing or limiting childbearing has essentialhealth benefits [1]. For instance, appropriate child spa-cing (i.e., 2 years or more) has been associated with a re-duced likelihood of preterm births, which is a keycontributor to neonatal and infant mortality [2]. Glo-bally, approximately 50 million women with either mis-timed or unwanted pregnancies have an inducedabortion, as a way of fertility control, most of which areunsafe resulting in high maternal mortality [3]. Thus,family planning (FP) through the use of modern contra-ceptives is of public health importance. However, gapsin terms of meeting the demand for FP services in devel-oping countries exist. Approximately 230 million womenin developing countries had an unmet need for modernFP methods in 2019 [4], an increase from the 225 mil-lion estimates reported in 2014 [5].Women are considered to have an unmet need for FP

if they want to stop or delay/postpone childbearing butare not using any method of contraception [5]. Unmetneed for FP is an important indicator for the gap interms of women’s reproductive intentions and theircontraceptive behavior [6]. Additionally, the unmet needfor FP is important for the assessment of the progresstowards achieving universal access to sexual and repro-ductive health services [7]. There are two types of unmetneed; unmet need for spacing and limiting. Unmet needfor spacing refers to a situation where a woman wants topostpone/delay pregnancy while limiting is when thewoman wants no more children and is not using anycontraception [8].Malawi has made tremendous efforts to improve the

accessibility of FP services by investing in human re-sources and training, deployment of lower cadres ofhealth professionals to provide community services andexpanding mobile and outreach services [9]. Therefore,commendable strides have been made in reducing un-met need from 35.0% in 1992 to 26.0% in 2010 [10].However, the 26.0% prevalence of unmet need in Malawiis relatively higher compared to the prevalence of unmetneed in Rwanda (19.0%) reported in the same year (i.e.,2010) [11]. Moreover, compared with Nigeria’s 10.8%rate of unintended pregnancies, a relatively high rate(47.0%) of unintended pregnancies in Malawi has beenreported thus underscoring that unmet need for FP maybe a persistent problem among Malawian women [12].Additionally, Malawi’s maternal mortality rate is one ofthe highest in sub-Saharan Africa at 510 deaths per 100,000 live births. Some of the leading causes of maternalmortality include unsafe abortions resulting from un-wanted pregnancies, high fertility rates and teenagepregnancies [13]. Therefore, improving access to modernFP methods is crucial in Malawi.

Previous studies in other countries have assessed fac-tors associated with unmet need for FP such as maternaleducational level, and maternal age [14]. Additionally,other studies reported that discussing FP issues with ahusband, and receiving partner support reduced the like-lihood of having unmet need for FP [15, 16]. An Ethiop-ian study revealed that unemployed women were morelikely to have unmet need for FP than their employedcounterparts [17]. However, inconsistent findings havebeen reported in different settings suggesting the needfor setting-specific data on the factors associated withunmet need. For example, while the area of residencewas associated with total unmet need in Burundi [14],no significant association was observed in Ghana [18].Additionally, there is a paucity of data on the contextualinfluences on unmet need. Community characteristicsinfluence the access to, and the utilization of health ser-vices [19]. Considering the effects of community charac-teristics may help account for the differences observedin literature regarding the factors associated with unmetneed for FP. Results may also help public health practi-tioners working in FP programs to design tailored FPinterventions.Therefore, the objective of this study was to investigate

individual- and community-level factors associated withunmet need in Malawian women, utilizing data from anationally representative sample.

MethodsStudy designThis was a cross-sectional study conducted using sec-ondary data from the 2015–16 Malawi demographichealth survey (MDHS). A detailed explanation of themethodology of the MDHS has been outlined elsewhere[20]. In brief, the survey used a two-stage cluster sam-pling method in which the first stage, clusters were ran-domly selected from the sampling frame (i.e. the 2008Malawi population and housing census) and householdlisting. The second stage involved a systematic selectionof households from the selected clusters.

Study settingMalawi is located in southern-central Africa and has apopulation of approximately 17.5 million people [21].Precisely, there has been a 35.0% population growth ratesince the last census was conducted in 2008 with thepopulation expected to double by 2042 (based on thecurrent annual growth rate of 2.9) [21]. As of 2018, amajority (84.0%) of Malawians were rural dwellers. Interms of contraceptive use, the use of modern contra-ceptive methods among married women of reproductiveage in Malawi increased from 28, 58, and 59.2% in 2004,2015, and 2016, respectively [21, 22], with the sustaineduse of injectable contraceptives and of long-acting and

Nkoka et al. BMC Public Health (2020) 20:705 Page 2 of 12

permanent methods of contraception. The observed in-crease has been attributed to high level commitment,continued health financing, expanded and innovativeservice delivery options [10]. Unmet need is high inMalawi, with 22% of married women—and 52% of un-married but sexually active women—aged 15–19 havingan unmet need for family planning [23]. Malawi has cre-ated an enabling policy environment to increase theutilization of FP services with a special focus on adoles-cent women. So far, the following achievements havebeen made regarding FP policies in Malawi: (1) the es-tablishment of national FP-related policies (such as Na-tional Reproductive Health Service Delivery Guidelines[24], National Sexual and Reproductive Health andRights Policy [25], and the National Population Policy[26]); (2) ensuring that these policies address some ofthe barriers to accessing FP services such as age or mari-tal status restrictions; (3) harnessing the prevention ofrights violations and practices that have a broad harmfuleffect on vulnerable groups (e.g., adolescent women); (4)integrating the FP policies with other youth-related pol-icies such as the National Youth Policy [27]; (5) address-ing the contextual factors that influence adolescentaccess to information and services; (6) and ensuring con-sistent implementation of the FP policies through the es-tablishment of accountability and necessary datacollection mechanisms [23].Data collection and sample size.Face to face interviews using pre-tested questionnaires

were conducted by experienced and trained data collec-tors. Information on sociodemographic, health-relatedfactors and use of contraceptives was collected. To as-sess unmet need for FP, a number of questions relatedto fertility intentions were asked in the MDHS women’squestionnaire (e.g., “Are you pregnant now?”, “Whenyou got pregnant, did you want to get pregnant at thattime?”, “Would you like to have (a/another) child, orwould you prefer not to have any (more) children?”). De-tailed questions asked to calculate unmet need for FPcan be obtained from the MDHS report published else-where [20]. A total of 24,562 out of 25,146 eligiblewomen were interviewed, representing a 98.0% responserate. The current analysis was restricted to fecundwomen who were married/living with a partner or un-married but sexually active (n = 15, 931). Women whowere infecund, and sexually inactive were excluded fromthe final analysis.

MeasuresOutcome measureBased on the revised definition of unmet need for FP byBradley et al. [28], we calculated unmet need for spacingby coding “1” to; (a) women who were not using contra-ception and were pregnant or postpartum amenorrhoeic

(last period not returned since last live birth in the last 2years) but wanted current pregnancy/last birth later and,(b) women who were not pregnant or postpartum ame-norrhoeic who were not using contraception but reportedwanting to have a child in the next ≥2 years, or wanted tohave children but had undecided timing or those thatwere undecided if they wanted a child. Therefore, womenwho reported to have been using contraception for spa-cing or with no unmet need were coded as “0”.Similarly, unmet need for limiting was calculated by

coding “1” to; (a) pregnant women or postpartum ame-norrhoeic women (last period not returned since last livebirth in the last 2 years) but did not want current preg-nancy/last birth at all, and (b) women who were notpregnant or postpartum amenorrhoeic who were notusing any contraceptive but reported wanting no morechildren. Those that reported using contraception forlimiting or with no unmet need were coded “0”.Total unmet need was a dichotomous variable calcu-

lated by combining unmet need for spacing or limiting.Women with unmet need for spacing or limiting werecoded as “1” while those using contraception for spacingor limiting or with no unmet need were coded as “0”.

Independent variablesIndependent variables were assessed at two-levels; level1 included the individual-level variables while level 2consisted of community/contextual factors.Individual-level variables were considered based on

their importance in literature [8, 14, 18] and included;sociodemographic factors such as woman’s age in years(15–24, 25–34, ≥ 35), marital status (married, unmar-ried) wealth (calculated using principal component ana-lysis in which the scores obtained from ownership ofhousehold items were grouped as poor (lower 40%),middle (middle 20%), and rich (upper 40%)), employed(yes or no), residence (urban, rural), region (northern,central, southern), women’s educational level was de-fined as the level of schooling ever attended [20] (no for-mal education, primary, secondary and higher). Fertilityintention drivers such as the number of children everhad (0, 1, 2+), whether the women ever experienced thedeath of child (yes or no), religion was categorized as“Catholics”, “protestants”, and “Muslims and other”, andmedia exposure (those who reported reading newspa-pers, listening to the radio, or watching television at leastonce a week were coded as “yes” or otherwise as “no”).Perceived distance to the health facility, categorized as“problem” or “no problem”, was included an access-related factor.To analyze variables at community-level, aggregation

of four key sociodemographic factors and one access-related factor from individual-level to community-levelwas done. These factors were selected based on their

Nkoka et al. BMC Public Health (2020) 20:705 Page 3 of 12

importance in previous studies [29, 30]. A communitywas defined as the primary sampling unit (i.e., cluster) ofthe MDHS survey. Community wealth, employment,women’s education, partner education, and distance tohealth facility were defined as the proportion of rich,employed, women with any education, and women whoperceived distance to health facility as a problem withina community. For easy interpretation, the percentageswere categorized into three levels using tertiles (low,middle, and high).

Statistical analysisDistribution of study participants’ characteristicsAll analyses were performed using Stata version 15.0(Stata Corp LP, College Station, TX, USA). The “svy”command was used to take into account the samplingweights and adjust for clustering effects of the hierarch-ical nature of the MDHS data. Distribution of study par-ticipants’ characteristics according to their unmet needstatus was assessed using Chi square tests. The total de-mand for FP was calculated by adding the percentage ofwomen with unmet need and the percentage of thoseusing any contraception method.

Modeling approachesThree modelling approaches were adopted namely; fixed ef-fect, random effects, and the goodness of fit assessments.

Fixed effects First, a two-level multilevel binary logisticregression model was fitted, using “xtmelogit” commandin Stata, to assess the association of individual- andcommunity-level factors, and the total unmet need.Women (level 1) were clustered within their communi-ties (level 2). Four models were tested; a null modelwhich was the unconditional model included the out-come variable(s) only to assess the variance in unmetneed between communities, model I included outcomeand individual-level variables, model II included out-come and community-level variables, and model III in-cluded the outcome variables, and both individual- andcommunity- level variables. The fixed effects for themultilevel binary logistic regression model were reportedas adjusted odds ratios (aORs) with 95% confidence in-tervals (CI).

Random effects The “xtmelogit” command allowed forthe assessment of random effects at the communitylevel. Measures of variation (random effects) wereassessed using several indicators such as area variance(AV) with 95% CI, the intraclass correlation coefficient(ICC), proportional change in variance (PCV), and themedian odds ratio (MOR).

Goodness of fit The goodness-of-fit of each model wasassessed using the Akaike information criterion (AIC),with a lower value representing a closer model fit. Thevariance inflation factor (VIF) was used to assess multi-collinearity. None of the variables displayed multicolli-nearity problems (all VIF < 10) (Table S1).

Sensitivity analysesIn sensitivity analyses, we repeated the main analyses byexcluding subsamples of the study population to examinethe effect on our results. First, women that were not usingcontraceptives but were pregnant or postpartum amenor-rhoeic and reported to have wanted current pregnancy/last birth, or those that were not using any contraceptionand were not pregnant or postpartum amenorrhoeic butreported to have wanted a child within 2 years wereregarded as having “no unmet need”. These were then ex-cluded from the analyses. Second, we repeated the analysisexcluding those that were using traditional methods to as-sess the factors associated with unmet need for modernFP methods. Third, the analysis was restricted to marriedwomen to control for factors such as age at first marriageand partner’s educational level.

Ethics statementPermission to utilize the data was obtained from thedemographic health survey program. The survey proto-col was reviewed and approved by the National HealthSciences Research Board of Malawi, Institutional ReviewBoard (IRB) of ICF Macro, and Centers for Disease Con-trol and Prevention (CDC) in Atlanta. Informed consentwas obtained at the beginning of each interview by theMDHS data collectors.

ResultsData for 15, 931 sexually active women (15, 110 marriedand 821 unmarried) (level 1) nested within 850 commu-nities (level 2) were analyzed. The overall prevalence ofunmet need among the total sample was 21.0% (3350).The prevalence of unmet need for spacing and limitingwere 12.6 and 8.4%, respectively.Specifically, among married women, the prevalence of

unmet need for FP was 18.7% while for contraceptiveuse was 59.2% (58.1% for modern and 1.1% for trad-itional methods). The total demand for FP was 77.9%while the proportion of demand satisfied by modernmethods was 74.6% (Fig. 1).Among unmarried sexually active women, the preva-

lence of contraceptive use was 39.8% (43.2% for modernand 1.2% for traditional methods). The total demand forFP was 84.2% while the proportion of demand satisfiedby modern methods was 51.3% (Fig. 2).

Nkoka et al. BMC Public Health (2020) 20:705 Page 4 of 12

Distribution of participants’ characteristics according tounmet need statusThe distribution of study participants’ characteristics ac-cording to unmet need (dichotomous variable) are listedin Table 1. Higher proportions of women with unmetneed were observed among women who were younger(22.1%), with no child (29.6%), from poor households(27.1%), unemployed (25.2%), unmarried (41.1%), fromrural areas (21.4%), those who belonged to Islam andother religions (22.5%), who had no media exposure(22.6%), from communities with a low percentage ofemployed women (24.5%), from communities with a lowpercentage of educated women (23.5).

Modeling approaches (fixed effects)Factors associated with total unmet needTable 2 displays the measures of association from themultilevel binary logistic regression model. Results from

model III which accounted for both individual- andcommunity- level variables revealed that married women[aOR = 0.41, 95% CI = 0.35–0.48], employed [aOR = 0.78,95% CI = 0.71–0.85], from the central and southern re-gions [aOR = 0.69, 95% CI = 0.60–0.80 and aOR = 0.80,95% CI = 0.70–0.92, respectively], with media exposure[aOR = 0.82, 95% CI = 0.75–0.90], from communitieswith a high percentage of women from rich households[aOR = 0.81, 95% CI = 0.67–0.96], from communitieswith a high percentage of employed women [aOR = 0.88,95% CI = 0.78–0.99], from communities with a middleand high percentage of educated women [aOR = 0.86,95% CI = 0.76–0.96 and aOR = 0.81, 95% CI = 0.70–0.93,respectively] were less likely to have total unmet needfor FP compared with their defined counterparts.Conversely, women aged ≥35 years were more likely[aOR = 1.19, 95% CI = 1.04–1.35] to have total unmetneed for FP compared with those aged 15–24 years.

Fig. 1 Unmet need among married women in Malawi, 2015–16 (Bradley et al. [28])

Nkoka et al. BMC Public Health (2020) 20:705 Page 5 of 12

Additionally, compared with Catholic women, womenbelonging to Islam and other religions were more likely[aOR = 1.12, 95% CI = 1.01–1.25] to have total unmetneed for FP. Women from communities with a middlepercentage of women complaining about the distance tohealth facility were more likely [aOR = 1.17, 95% CI =1.03–1.32] to have unmet need for FP compared withthose from communities with a low percentage ofwomen complaining about the distance to health facility.

Modeling approaches (random effects)Measures of variation for the total unmet need outcomesare displayed in Table 2. In the null models, the use ofmultilevel modeling was justified by the significant vari-ation in total unmet need (σ2 = 0.14, 95% CI 0.10–0.20).The ICC for the total unmet need was 4.0% suggestingthat variation in unmet need status may be attributableto other unobserved community characteristics. Thefinal model revealed significant variances and the MORof 1.33 showed the effects of community heterogeneity(i.e., suggesting that if a married woman moved to acommunity with a higher probability of total unmetneed, the median increase in the odds of having totalunmet need for FP would be 1.33-fold). Additionally,35.7% of the variance in the odds of having total unmetneed across communities explained by both individual-and community-level factors, as indicated by the PCV.

Sensitivity analysesAfter excluding those categorized as having “no unmetneed”, similar results to those when these were includedin the analyses were observed (Table S2). Additionally,there were 188 (1.2%) women that reported using trad-itional contraceptive methods. After excluding thoseusing traditional contraceptive methods from analyses,the results were fairly consistent as those when thesewomen were included in the total sample (Table S3).Similarly, when the analysis was restricted to marriedwomen only to control for marriage-related factors, nosubstantial differences from the main results except thateducational level was no longer associated with unmetneed (Table S4).

DiscussionThis study examined the individual- and community-level factors associated with unmet need for FP inMalawi. Apart from significant individual-level factorsassociated with unmet need for FP, the study also re-vealed significant community effects. Specifically, womenfrom communities with a high percentage of womenfrom rich households and from communities with a highpercentage of educated women exhibited the same re-duction (19.0%) in the likelihood of having total unmetneed compared with those from communities with lowpercentages of women from rich households and edu-cated women, respectively.

Fig. 2 Unmet need among unmarried 2015–16 (Bradley et al. [28])

Nkoka et al. BMC Public Health (2020) 20:705 Page 6 of 12

Table 1 Distribution of study participants according to unmet need status among sexually active women

Variable Unmet need – (n = 15,931)

No (n = 12,581) Yes (n = 3350) p-value a

Individual-level factors

Woman’s age (years) 0.011

15–24 4123 (77.9) 1172 (22.1)

25–34 5014 (80.6) 1209 (19.4)

≥ 35 3444 (78.0) 969 (21.9)

Number of children ever had < 0.001

0 348 (70.4) 357 (29.6)

1 2221 (80.3) 546 (19.7)

2+ 9512 (79.5) 2447 (20.5)

Wealth 0.049

Poor 4918 (77.9) 1397 (27.1)

Middle 2432 (79.0) 649 (21.0)

Rich 5230 (80.0) 1304 (20.0)

Marital status < 0.001

Unmarried 483 (58.9) 338 (41.1)

Married 12,098 (80.1) 3012 (19.9)

Employed < 0.001

No 3342 (74.7) 1129 (25.2)

Yes 9239 (80.6) 2221 (19.4)

Residence 0.046

Urban 2195 (80.9) 518 (19.1)

Rural 10,386 (78.6) 2832 (21.4)

Region < 0.001

Northern 1473 (75.8) 471 (24.2)

Central 5691 (82.1) 1239 (17.9)

Southern 5417 (76.8) 1640 (23.2)

Woman’s educational level 0.216

No formal education 1571 (77.0) 470 (23.0)

Primary 8108 (79.2) 2135 (20.8)

Secondary and higher 2902 (79.6) 745 (20.4)

Religion 0.001

Catholics 2556 (81.1) 596 (18.9)

Protestants 3018 (80.8) 717 (19.2)

Muslims and other 7007 (77.5) 2037 (22.5)

Experienced death of child < 0.011

No 11,642 (79.7) 2968 (20.3)

Yes 939 (71.1) 382 (28.9)

Media exposure < 0.001

No 7579 (77.4) 2213 (22.6)

Yes 5002 (81.5) 1137 (18.5)

Distance to HF 0.084

No problem 5610 (79.8) 1421 (20.2)

Problem 6971 (78.3) 1929 (21.7)

Nkoka et al. BMC Public Health (2020) 20:705 Page 7 of 12

The overall prevalence for unmet need was 21.0% (ahigher prevalence of unmet need for spacing (12.6%)was observed compared with 8.4% for unmet need forlimiting). Among married women, the total unmet needwas 18.7% representing a decline from the 26.0% unmetneed prevalence among married women reported inMalawi in 2010. This reduction in unmet need inMalawi over the years could be moderately attributed tothe youth-friendly FP services (under the youth friendlyhealth services program) which has aimed at improvingthe usage of modern contraceptive methods among theyouth. However, disparities in access to FP methods hasbeen reported among different sociodemographic strataand understanding factors associated with unmet needin Malawi is vital for public health practitioners to de-sign targeted, and strengthen the already existing,interventions.

Factors associated with total unmet needIn the current analysis, married women were less likelyto have total unmet need compared with unmarriedwomen. As observed, the demand for FP was higheramong sexually unmarried women compared to marriedwomen (Fig. 1 and Fig. 2). The cultural opposition to be-ing pregnant while unmarried may raise the need for FPservices among sexually active unmarried women [31].Consistent with a study from Ghana [18], employedwomen were less likely to have total unmet need.Employed women may have better access to quality

health services as they may be able to afford privatehealth insurance compared with their unemployed coun-terparts [32]. Additionally, employed women are morelikely to be independent and have better autonomy ontheir health decisions and therefore, may exhibit betterbehaviors in health services utilization (including contra-ceptive use) [32, 33].Regional variations in terms of the total unmet need

were observed with women from the central and south-ern region being less likely to have unmet need for FP. Ithas been reported that in Malawi, women from thenorthern region have the lowest rates of use of moderncontraceptives [20]. The northern region is dominantly apatrilineal society hence most women depend on theirpartners when it comes to healthcare decisions (includ-ing FP). A Ugandan study reported that men were lesslikely to have knowledge of contraceptives with most ofthem expressing fear of the side effects of moderncontraceptive methods to their partners [34]. Addition-ally, women in the Northern region were reported tohave more co-wives compared with those from the cen-tral and southern regions [20]. A 2013 study in NorthernMalawi revealed polygamy to be a driver of fertility pref-erence disagreements which may in turn, influence un-met need for FP in this region [35]. Lastly, a 2012longitudinal study conducted in Northern Malawi re-ported a high discontinuation of contraceptive useamong women [36] which may highlight the effects ofunderlying institutional-based factors such as stock-outs.

Table 1 Distribution of study participants according to unmet need status among sexually active women (Continued)

Variable Unmet need – (n = 15,931)

No (n = 12,581) Yes (n = 3350) p-value a

Community-level factors

Community wealth 0.051

Low 5063 (77.9) 1433 (22.1)

Middle 4158 (78.7) 1124 (21.3)

High 3360 (80.9) 793 (19.1)

Community employment < 0.001

Low 3631 (75.5) 1179 (24.5)

Middle 4075 (79.4) 1058 (20.6)

High 4875 (81.4) 1113 (18.6)

Community women’s education < 0.001

Low 4435 (76.5) 1365 (23.5)

Middle 4557 (80.2) 1122 (19.8)

High 3589 (80.6) 863 (19.4)

Community distance to HF 0.115

Low 3263 (80.7) 779 (19.3)

Middle 4889 (78.2) 1366 (21.8)

High 4429 (78.6) 1205 (21.4)ap-value from chi square test, bold means significant at p < 0.05

Nkoka et al. BMC Public Health (2020) 20:705 Page 8 of 12

In the current analysis, women who reported to havehad exposure to media at least once a week were lesslikely to report having the total unmet need. A study in

Table 2 Multilevel logistic analysis of factors associated withtotal unmet need among sexually active womenVariable Null model Model I

aOR(95% CI)

Model IIaOR(95% CI)

Model IIIaOR(95% CI)

Individual-level factors

Woman’s age (years)

15–24 1.00 1.00

25–34 0.98(0.87–1.10)

0.98(0.87–1.10)

≥ 35 1.17(1.03–1.33)

1.19(1.04–1.35)

Number of children ever had

0 1.00 1.00

1 0.61(0.37–1.01)

0.60(0.36–1.00)

2+ 0.62(0.37–1.04)

0.61(0.36–1.03)

Wealth

Poor 1.00 1.00

Middle 0.94 (0.84–1.05)

0.95(0.85–1.06)

Rich 0.94 (0.84–1.05)

0.99(0.88–1.11)

Marital status

Unmarried 1.00 1.00

Married 0.42(0.35–0.49)

0.41(0.35–0.48)

Employed

No 1.00 1.00

Yes 0.76(0.69–0.82)

0.78(0.71–0.85)

Residence

Urban 1.00 1.00

Rural 1.09(0.96–1.25)

0.91(0.77–1.07)

Region

Northern 1.00 1.00

Central 0.78(0.69–0.89)

0.69(0.60–0.80)

Southern 0.91(0.81–1.03)

0.80(0.70–0.92)

Woman’s educational level

No formaleducation

1.00 1.00

Primary 0.85(0.75–0.96)

0.90(0.79–1.02)

Secondary andhigher

0.87(0.74–1.02)

0.94(0.80–1.11)

Religion

Catholics 1.00 1.00

Protestants 1.01(0.89–1.14)

1.00(0.88–1.13)

Muslims andother

1.14(1.03–1.27)

1.12(1.01–1.25)

Experienced death of child

No 1.00 1.00

Table 2 Multilevel logistic analysis of factors associated withtotal unmet need among sexually active women (Continued)Variable Null model Model I

aOR(95% CI)

Model IIaOR(95% CI)

Model IIIaOR(95% CI)

Yes 0.90(0.55–1.47)

0.88(0.54–1.44)

Media exposure

No 1.00 1.00

Yes 0.82(0.75–0.90)

0.82(0.75–0.90)

Distance to HF

No problem 1.00 1.00

Problem 1.09(1.01–1.19)

1.05(0.96–1.15)

Community-level factors

Community wealth

Low 1.00 1.00

Middle 0.94(0.84–1.05)

0.90(0.81–1.01)

High 0.87(0.76–1.01)

0.81(0.67–0.96)

Community employment

Low 1.00 1.00

Middle 0.87 (0.78–0.97)

0.95(0.85–1.06)

High 0.76 (0.67–0.85)

0.88(0.78–0.99)

Community women’s education

Low 1.00 1.00

Middle 0.86 (0.77–0.98)

0.86(0.76–0.96)

High 0.84 (0.74–0.95)

0.81(0.70–0.93)

Community distance to HF

Low 1.00 1.00

Middle 1.18 (1.04–1.33)

1.17(1.03–1.32)

High 1.11(0.97–1.27)

1.08(0.93–1.24)

Measures of variation

Area variance(95% CI)

0.14 (0.10–0.19)

0.10(0.06–0.15)

0.11(0.07–0.16)

0.09(0.05–0.14)

ICC (%) 4.0 2.9 3.2 2.7

PCV (%) Ref. 28.6 21.4 35.7

MOR 1.43 1.35 1.37 1.33

Model Fit statistic

AIC 16,154.26 15,877.07 16,116.63 15,860.86

Null model contains no explanatory variables; Model I includes individual-levelfactors only; Model II includes community-level factors only; Model III includesboth individual-level and community-level factorsaOR adjusted odds ratio, CI confidence internal, ICC intraclass correlationcoefficient, MOR median odds ratio, PVC proportional change in variance, AICAkaike information criterion

Nkoka et al. BMC Public Health (2020) 20:705 Page 9 of 12

Botswana reveled that women who reported to havebeen listening to radio at least once a week were lesslikely to have unmet need [37]. Strengthening mediaprograms in disseminating FP messages on the import-ance of FP methods, and where they can be accessed isthus essential in Malawi.Our findings revealed that women who belonged to

Islam and other religions were more likely to have totalunmet need for FP compared with those belonging tothe Catholic religion. This is consistent with a Nepalesestudy in which Muslim women were positively associ-ated with having unmet need [38]. Religious beliefs havebeen shown to influence health behaviors. For instance,religiosity influenced fertility preferences in the UnitedStates [39]. In Iran, fertility preferences were higheramong individuals with stronger religious beliefs [40].The variations within different religious groups observedin this study underscore the need for FP programs to en-gage different religious institutions and influential reli-gious leaders to effectively scale up FP services.The effects of community characteristics on health

outcomes and behaviors have been well-documented[32, 41]. Findings from the current analysis revealed thatwomen from communities with a high percentage ofwomen from rich households and educated women wereless likely to have total unmet need. Educated womenare more likely to comprehend health messages and de-mand services [42]. Additionally, educated women aremore likely to be empowered which may subsequentlyincrease their contraceptive use [43, 44]. Similarly,women from rich households have better chances ofaccessing information and affording private health facil-ities to access FP services. As such, women from com-munities with a high percentage of rich and educatedwomen may learn from others on the importance ofusing FP services and where these may be accessed.Compared with women from communities with a lowpercentage of women complaining of the distance to ahealth facility, those from communities with a middlepercentage of women complaining of the distance tohealth facility were more likely to have unmet need forFP. In Ethiopia, proximity to a health facility was inde-pendently associated with contraceptive utilization [45].The associations with perceived distance were more pro-nounced at the community level than at the individuallevel because in the local setting, individual effects maybe attenuated by services that have been rolled out (i.e.,mobile clinic services and community health workers)that mainly target individuals [9].

Policy/program implicationsFirst, regional differences were observed. FP programsand interventions need to be strengthened in the north-ern region of Malawi. Second, as observed, employed

women were less likely to have total unmet need sug-gesting that empowering women may go a long way inaddressing FP challenges. Third, there were unobservedor unmeasured community factors that influenced un-met need for FP. This highlights that there are factors op-erating at the community level, not included in thecurrent analysis, which may be associated with unmetneed in Malawi. These may include but are not limited tocultural differences between communities (that may ul-timately influence misconceptions and myths about FP),and community outreach, engagement, and mobilizationefforts. Therefore, FP programs need to conduct thoroughcommunity profiling, and strengthen their community en-gagement approaches involving relevant stakeholders suchas community leaders and religious institutions.

Strengths and limitationsThe study included a nationally representative sample ofwomen in Malawi therefore, results from the currentanalysis may be generalized to Malawian women. Thehierarchical nature of the DHS dataset allowed for ex-ploration of community effects which may have an influ-ence on FP programming in Malawi. A wide range offactors were assessed in this study to strengthen the as-sociations observed. The cross-sectional nature of thestudy means causality cannot be inferred. The use of ad-ministratively defined boundaries has the potential ofintroducing misclassification for unfitted administrativecommunities.

ConclusionA higher rate of unmet need for spacing (12.6.8%) wasobserved compared to the rate for unmet need for limit-ing (8.4%). In Malawi, factors influencing unmet needfor FP operate at both individual and community level.FP programs in Malawi should be strengthened in disad-vantaged communities, and the northern region. Quali-tative research is needed in Malawi to understand someof the observations made in the current analysis and todivulge more information on the influences of cultural,and religious beliefs that may explain some of the un-accounted community effects.

Supplementary informationSupplementary information accompanies this paper at https://doi.org/10.1186/s12889-020-08885-1.

Additional file 1. The supplement has tables s1 (multicollinearity tests),and s3- s4 (sensitivity analyses)

AbbreviationsaOR: Adjusted odds ratio; CI: confidence interval; MDHS: Malawidemographic health survey; FP: Family planning; ICC: Intraclass correlationcoefficient; PVC: Proportional change in variance; AIC: Akaike informationcriterion; MOR: Median odds ratio

Nkoka et al. BMC Public Health (2020) 20:705 Page 10 of 12

AcknowledgementsWe acknowledge the International Classification of Functioning Disability andHealth (ICF) for the permission to use the MDHS data set for analysis.

Authors’ contributionsON and WMM designed the study. ON applied for data access, conductedthe data analyses, and drafted the manuscript. PAMN, EBM, VK and WMMassisted in literature review, provided advice in data analysis, as well asmanuscript review for intellectual content. All authors read and approvedthe final manuscript.

FundingThis study did not receive any funding.

Availability of data and materialsThe study used, with permission, data from the International Classification ofFunctioning, Disability, and Health (ICF). The data is publicly available uponrequest from the ICF on (https://dhsprogram.com/data/available-datasets.cfm).

Ethics approval and consent to participateBefore each interview was conducted, a verbal informed consent was soughtby each interviewer reading a prescribed statement to the respondent andrecording in the questionnaire whether or not the respondent consented (orprovided assent on behalf of minors). Then the interviewer signed his or hername attesting to the fact that he/she read the consent statement to therespondent. The method of collecting consent has been standardized in theDHS survey to ensure consistency as some participants are not able to write.The Malawi’s National Health Sciences Research Committee and theInstitutional Review Board (IRB) of ICF Macro, and the Centers for DiseaseControl (CDC) in Atlanta approved the protocol for the 2015–2016 MDHS.

Consent for publicationNot applicable.

Competing interestsThe authors declare that they have no competing interests.

Author details1Institute for Health Research and Communication (IHRC), P.O Box 1958,Lilongwe, Malawi. 2School of Public Health, College of Public Health, TaipeiMedical University, 250 Wuxing Street, Xinyi Taipei, Taiwan 110. 3Ministry ofHealth, P.O. Box 30377, Lilongwe 3, Malawi. 4University of Malawi, College ofMedicine, Malaria Alert Centre, Private Bag 360, Chichiri, Blantyre 3, Malawi.

Received: 2 October 2019 Accepted: 10 May 2020

References1. Williamson LM, Parkes A, Wight D, Petticrew M, Hart GJ. Limits to modern

contraceptive use among young women in developing countries: asystematic review of qualitative research. Reprod Health. 2009;6:3–3.

2. Organization WH: Neonatal and perinatal mortality: country, regional andglobal estimates. 2006. Available from: https://apps.who.int/iris/bitstream/handle/10665/43444/9241563206_eng.pdf;jsessionid=37D57187D78DEF7F0DC7F53E26906427?sequence=1. Accessed 17 April 2020.

3. Ganatra B, Gerdts C, Rossier C, Johnson BR Jr, Tunçalp Ö, et al. Global, regional,and subregional classification of abortions by safety, 2010–14: estimates from aBayesian hierarchical model. Lancet. 2017;390(10110):2372–81.

4. United Nations Department of Economic and Social Affairs (2019). Estimatesand projections of family planning indicators. Available from: https://www.un.org/en/development/desa/population/theme/family-planning/cp_model.asp. Accessed 17 April 2020.

5. Sedgh G, Ashford LS, Hussain R. Unmet need for contraception indeveloping countries: examining Women’s reasons for not using a method.2 New York: Guttmacher Institute; 2016. 2:2015–2016. Available from:https://www.guttmacher.org/report/unmet-need-for-contraception-in-developing-countries. Accessed 17 April 2020.

6. Bradley SEK, Casterline JB. Understanding unmet need: history, theory, andmeasurement. Stud Fam Plan. 2014;45(2):123–50.

7. Lince-Deroche N, Pleaner M, Morroni C, Mullick S, Firnhaber C, Harries J,et al. Achieving universal access to sexual and reproductive health services:the potential and pitfalls for contraceptive services in South Africa. AfrHealth Rev. 2016;2016(1):95–108.

8. Wulifan JK, Brenner S, Jahn A, De Allegri M. A scoping review ondeterminants of unmet need for family planning among women ofreproductive age in low and middle income countries. BMC WomensHealth. 2016;16:2–2.

9. USAID. Project R: Making family planning acceptable, accessible, andaffordable: The experience of Malawi. 2012. Available from: http://www.respond-project.org/archive/files/1/1.1/1.1.1/Brief6-2012-Making-Family-Planning.pdf. Accessed 17 April 2020.

10. National Statistical Office - NSO/Malawi and ICF Macro. Malawidemographic and health survey 2010. Zomba, Malawi: NSO/Malawi and ICFMacro; 2011.

11. Basinga P, Moore AM, Singh S, Remez L, Birungi F, Nyirazinyoye L.Unintended pregnancy and induced abortion in Rwanda: causes andconsequences. Guttmacher Institute: New York; 2012. Available from:https://www.guttmacher.org/report/unintended-pregnancy-and-induced-abortion-rwanda. Accessed 17 April 2020.

12. Ameyaw EK, Budu E, Sambah F, Baatiema L, Appiah F, Seidu A-A, AhinkorahBO. Prevalence and determinants of unintended pregnancy in sub-SaharanAfrica: a multi-country analysis of demographic and health surveys. PLoSOne. 2019;14(8):e0220970.

13. Haddad LB, Nour NM. Unsafe abortion: unnecessary maternal mortality. RevObstet Gynecol. 2009;2(2):122–6.

14. Nzokirishaka A, Itua I. Determinants of unmet need for family planningamong married women of reproductive age in Burundi: a cross-sectionalstudy. Contracept Reprod Med. 2018;3:11.

15. Worku SA, Ahmed SM, Mulushewa TF. Unmet need for family planning andits associated factor among women of reproductive age in Debre Berhantown, Amhara, Ethiopia. BMC Res Notes. 2019;12(1):143.

16. Shifa GT, Kondale M. High unmet need for family planning and factorscontributing to it in southern Ethiopia: a community based cross-sectionalstudy. Glob. J. Med. Res; 2014.

17. Genet E, Abeje G, Ejigu T. Determinants of unmet need for family planningamong currently married women in Dangila town administration, Awi zone,Amhara regional state; a cross sectional study. Reprod Health. 2015;12:42.

18. Wulifan JK, Mazalale J, Kambala C, Angko W, Asante J, Kpinpuo S, et al.Prevalence and determinants of unmet need for family planning amongmarried women in Ghana-a multinomial logistic regression analysis of theGDHS, 2014. Contracept Reprod Med. 2019;4:2–2.

19. Hillemeier MM, Lynch J, Harper S, Casper M. Measuring contextual characteristics forcommunity health. Health Serv Res. 2003;38(6 Pt 2):1645–717.

20. National Statistical Office - NSO/Malawi and ICF. Malawi demographic andhealth survey 2015–16. Zomba, Malawi: NSO and ICF; 2017.

21. Goverment, Malawi. Malawi: 2018 Population and Housing Census. Zomba:National Statistical Office; 2018. http://populationmalawi.org/wp1/wp-content/uploads/2019/10/2018-Malawi-Population-and-Housing-Census-Main-Report-1.pdf. Accessed 17 Apr 2020.

22. World Bank . Trading Economics: Malawi–Contraceptive Prevalence (% ofWomen Ages 15–49). 2016. Available from https://tradingeconomics.com/malawi/contraceptive-prevalence-percent-of-women-ages-15-49-wb-data.html. Accesed 17 April 2020.

23. USAID & Policy Plus. Review of Adolescent Family Planning Policies inMalawi. 2017. Available from: http://www.healthpolicyplus.com/ns/pubs/7159-7279_MalawiAdolescentFPPolicyBrief.pdf. Accessed 17 April 2020.

24. Ministry of Health and Population. Malawi National Reproductive HealthServices Delivery Guidelines 2014–2019. Availble from: http://lifesavingcommodities.org/wp-content/uploads/2015/03/Malawi-2014_National-Reproductive-Health-Service-Delivery-Guidelines-2014-2019_Guidelines.pdf. Accessed April 17 2020.

25. Ministry of Health and Population. National sexual and reproductive healthrights (SRHR) policy (2017–2022). Available from: https://malawi.unfpa.org/sites/default/files/resource-pdf/Malawi_National_SRHR_Policy_2017-2022_16Nov17.pdf. Accessed 17 April 2020.

26. Ministry of Economic Planning and Development. National PopulationPolicy. Availed from: http://populationmalawi.org/wp1/wp-content/uploads/2014/04/Malawi-National-Population-Policy-2012.pdf. Accessed 17April 2020.

Nkoka et al. BMC Public Health (2020) 20:705 Page 11 of 12

27. Ministry of Youth and Sports. National Youth Policy. Available from: https://cepa.rmportal.net/Library/government-publications/National%20Youth%20Policy%202013.pdf/view. Accessed 17 April 2020.

28. Bradley SE, Croft TN, Fishel JD, Westoff CF: Revising unmet need for familyplanning. 2012. Available from: https://dhsprogram.com/pubs/pdf/AS25/AS25[12June2012].pdf. Accessed 17 April 2020.

29. Mutumba M, Wekesa E, Stephenson R. Community influences on moderncontraceptive use among young women in low and middle-incomecountries: a cross-sectional multi-country analysis. BMC Public Health. 2018;18(1):430.

30. Tiruneh FN, Chuang K-Y, Ntenda PAM, Chuang Y-C. Individual-level andcommunity-level determinants of cervical cancer screening among Kenyanwomen: a multilevel analysis of a Nationwide survey. BMC Womens Health.2017;17(1):109.

31. Dansereau E, Schaefer A, Hernández B, Nelson J, Palmisano E, Ríos-ZertucheD, Woldeab A, Zúñiga MP, Iriarte EM, Mokdad AH, et al. Perceptions of andbarriers to family planning services in the poorest regions of Chiapas,Mexico: a qualitative study of men, women, and adolescents. ReprodHealth. 2017;14(1):129.

32. Tiruneh FN, Chuang K-Y, Chuang Y-C. Women’s autonomy and maternalhealthcare service utilization in Ethiopia. BMC Health Serv Res. 2017;17(1):718.

33. Viswan SP, Ravindran TKS, Kandala N-B, Petzold MG, Fonn S. Sexual autonomyand contraceptive use among women in Nigeria: findings from thedemographic and health survey data. Int J Women’s Health. 2017;9:581–90.

34. Thummalachetty N, Mathur S, Mullinax M, DeCosta K, Nakyanjo N, Lutalo T,et al. Contraceptive knowledge, perceptions, and concerns among men inUganda. BMC Public Health. 2017;17(1):792.

35. Baschieri A, Cleland J, Floyd S, Dube A, Msona A, Molesworth A, Glynn JR,French N. Reproductive preferences and contraceptive use: a comparison ofmonogamous and polygamous couples in northern Malawi. J Biosoc Sci.2013;45(2):145–66.

36. Dasgupta ANZ, Zaba B, Crampin AC. Contraceptive dynamics in ruralnorthern Malawi: a prospective longitudinal study. Int Perspect Sex ReprodHealth. 2015;41(3):145–54.

37. Letamo G, Navaneetham K. Levels, trends and reasons for unmet need forfamily planning among married women in Botswana: a cross-sectionalstudy. BMJ Open. 2015;5(3):e006603.

38. Bhanderi MN, Kannan S. Untreated reproductive morbidities among evermarried women of slums of Rajkot City, Gujarat: the role of class, distance,provider attitudes, and perceived quality of care. J Urban Health. 2010;87(2):254–63.

39. Hayford SR, Morgan SP. Religiosity and fertility in the United States: the roleof fertility intentions. Soc Forces. 2008;86(3):1163–88.

40. Khadivzade T. Arghavani EJJoM, health R. religious beliefs and fertilitypreferences among engaged couples, referring to premarital counselingcenters of Mashhad, Iran. J Midwifery Reprod Health. 2014;2(4):238–45.

41. Nkoka O, Chuang T-W, Chen Y-H. Multilevel analysis of factors associatedwith treatment-seeking behaviors among caregivers with febrile children inMalawi. Am J Trop Med Hyg. 2019;100(6):1454–65.

42. Ross CE, Wu C-1. The links between education and health. Am Sociol Rev.1995;60(5):719–45.

43. Samarakoon S, Parinduri RA. Does education empower women? Evidencefrom Indonesia. World Dev. 2015;66:428–42.

44. Hameed W, Azmat SK, Ali M, Sheikh MI, Abbas G, Temmerman M, Avan BI.Women's empowerment and contraceptive use: the role of independentversus Couples' decision-making, from a lower middle income countryperspective. PLoS One. 2014;9(8):e104633.

45. Shiferaw S, Spigt M, Seme A, Amogne A, Skrøvseth S, Desta S, Radloff S, TsuiA, GeertJan D. Does proximity of women to facilities with better choice ofcontraceptives affect their contraceptive utilization in rural Ethiopia? PLoSOne. 2017;12(11):e0187311.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Nkoka et al. BMC Public Health (2020) 20:705 Page 12 of 12